Routinely measured cations and anions include sodium, potassium, chloride and bicarbonate. The sum of all circulating cations must equal the sum of all circulation anions [1
]. The AG exists because the total unmeasured anions exceed the total unmeasured cations. Population studies using contemporary measurements of serum electrolytes suggest the normal range of AG to be between six and ten [1
]. Furthermore, the concentration of potassium in the blood is usually negligible compared to that of sodium, chloride and bicarbonate. Therefore, the AG equals the sum of sodium ions minus the sum of the chloride ions and bicarbonate ions.
A low AG or a negative AG is uncommon in clinical medicine. There are very few clinical conditions associated with a low AG. Additionally, laboratory error may be a possible cause of a low AG finding. Thus, the test was repeated in our patient. The AG was still between zero and minus one. Normally, the majority of the serum AG is due to the sum of the anionic charges on circulating proteins. Albumin being the most abundant of all circulating proteins, hypoalbuminemia can result in a low AG. For each 1 g/dL decrement in the serum concentration of albumin, the serum AG decreases by 2.5 mEq/L [1
]. In our patient, the serum albumin was 2.0 g/dL. It is therefore unlikely that this degree of hypoalbuminemia resulted in a zero or negative AG.
A rise in unmeasured cations including hyperkalemia, hypercalcemia and hypermagnesemia can also lower the AG. This usually remains clinically insignificant [1
]. Severe intoxication with bromide, iodide or lithium can also lower the serum AG [1
]. Our patient deteriorated four days after hospitalization and was thus unlikely to have been exposed to toxic levels of these substances.
In the evaluation of a verifiable low AG, one must also consider the presence of paraproteins. Overproduction of cationic monoclonal immunoglobulins, such as immunoglobulins G, can result in increased unmeasured cations. The serum AG in these patients can be below the normal range or, rarely, can even be negative [1
Our patient presented with what appeared to be a sickle cell crisis. This likely masked his underlying MM and delayed the diagnosis. There are rare reports of SCD coexisting with MM [6
]. The serum AG of zero led us to suspect and test for MM. Much of our patient's clinical presentation was initially thought to be caused by SCD. In retrospect, we must additionally consider the role of MM in his symptomatology. His refractory anemia can be explained by plasma cell infiltration of the bone marrow. The recurrent bone pain, thought to be secondary to a vaso- occlusive crisis, may be also explained by increased osteoclast activity, mediated particularly by interleukin 6 seen in MM [7
]. The increased frequency of infection was likely a combination of SCD-associated autosplenectomy and immunodeficiency caused by the hypogammaglobulinemia of MM. Interestingly, the altered mental status remains unexplained despite a detailed workup. His encephalopathy cleared with treatment of his MM. Although somewhat speculative, his increased ammonia may have originated from his monoclonal plasma cells even in the presence of normal liver function [8